640 Proceedings of the Royal Society of Edinburgh. [Sess. 
of the series made with No. 21. If we consider the series made with both 
hydrometers for which the variation of temperature was not greater than 
0'05°, the percentages are almost identical, namely, 89*5 for No. 17 and 89*2 
for No. 21. 
In the paper, a sensible variation of temperature is defined to be one 
that is perceptible on a thermometer divided into tenths of a Centigrade 
degree, when each such interval of temperature on the scale occupies a 
length of one millimetre on the stem of the thermometer. 
Table No. 2. — Giving Statistics of Constancy or Variability of 
Temperature during Experiments in the Earlier Years of 
Work in the Davy-Far ad ay Laboratory. 
Amplitude of Variation 
of the Temperature of 
the Liquid while a 
Series of Observations 
was being made with 
the Hydrometer. 
Number of Series of Observations so made 
with Hydrometer. 
No. 
17. 
No. 21. 
Absolute. 
Per cent. 
Absolute. 
Per cent. 
o-oo° 
568 
68-0 
265 
55-2 
•01° 
48 
5-7 
51 
106 
•02° 
68 
8T 
54 
11-3 
•03° 
19 
23 
23 
5-0 
•04° 
12 
1-4 
16 
3-3 
•05° 
34 
4-0 
18 
3*8 1 
*05° to T0° 
69 
8-3 
40 
8-4 
T0° to *20° 
16 
1-9 
11 
2-2 
•20° to -30° 
3 
0-3 
1 
0-2 
Totals 
837 
1000 
479 
100-0 
In order to obtain full advantage of the degree of precision of which 
the hydrometric method is capable, the experimenter must train himself 
in the management of the climate of his laboratory, so as to produce the 
constancy of temperature of which the numbers in this table are evidence. 
The hydrometer itself is only an instrument like the balance or the 
thermometer. It is true that all of them are necessary; but what is 
indispensable is the competence and the earnestness of the investigator 
who handles them. 
When the systematic research was begun in the year 1902 my first 
object w~as to study the specific gravity and displacement of equivalent 
solutions of the salts of the Ennead (K,Rb,Cs ; Cl,Br,I). The most concen- 
trated solution contained \ gram-molecule of the salt dissolved in 1000 
grams of water. Using hydrometers Nos. 17 and 21, it was necessary to 
operate with at least 1000 grams of the solution. In view of the rarity and 
very high price of the salts of caesium, it was impossible to work with 
solutions containing more than 1 gram-molecule. 
